MEMS Energy Harvesters Design and Simulation
Piezoelectric Vibratory Energy Harvester Design and Simulation
MEMS-based energy harvesters have the potential to provide “perpetual” power to small systems. Applications include medical, consumer, automotive and environmental devices. Wireless sensor networks (WSN) containing MEMS-based energy harvesters (Figure 1, below) are of particular interest for in-situ environmental, health and habit monitoring, where batteries are difficult or impractical to replace.
Figure 1: A typical WSN node (Yole Development, dMEMS Conference April 2012)
Experts believe that piezo-electric vibratory energy harvesting is one of the most promising MEMS-based approaches. Successful commercialization of these devices will depend upon maximizing energy capture at target environmental conditions.
Design Challenges
The design criteria for a piezoelectric energy harvester includes the frequency of operation, the power generated and the power transferred to the management circuit. The frequency of operation can be determined by running a modal analysis in a conventional finite element analysis (FEA) tool. The power generated and transferred, however, are highly dependent on the power management circuit and must be simulated in a closed-loop circuit. Thus, the design platform chosen must be able to simulate the coupled piezo-mechanics and the electronics simultaneously in order to be useful in energy harvester design.
New Hybrid Methodology
1. Rapid MEMS Design Exploration
Coventor’s design flow starts by constructing a parametric model of the piezoelectric vibratory harvester in MEMS+. The designer works in a 3D graphical environment to create a parametric model by assembling higher-order, MEMS-specific, finite elements (piezo-mechanical shells). Each element is linked to the MEMS process description and material database so that piezo-electric material properties and electrodes are assigned automatically. The higher-order elements provide a precise mathematical description of the device physics, and include the non-linear mechanics that are inherent in these devices. Furthermore, these higher order components have been specifically crafted to simulate extremely quickly within system and circuit modeling environments like MATLAB®, Simulink® and Cadence® Spectre®. Engineers working at the device level generally prefer to use MATLAB and Simulink, while engineers working on designing the IC and system prefer to work in the Cadence platform.
Using MEMS+ higher-order finite elements provides several benefits. First, design teams no longer need to spend valuable time hand crafting reduced-order models from FEA and/or analytical expressions. Second, MEMS+ models include non-linear effects, unlike hand-crafted models which are often linear only. Non-linear effects that may affect device performance are included in the MEMS+ model, reducing the chances of having to redesign the device during the final stages of development.
Figure 2: Coventor design flow for energy harvesters
2. Power Management and Circuit Design with Cadence
The MEMS+ model can be imported to the Cadence Virtuoso® environment with a few mouse clicks, and placed in a schematic that includes the conditioning circuit. The combined harvester and circuit can then be simulated in Cadence Spectre. Both the harvester parameters and circuit parameters can then be varied simultaneously to optimize device performance. For example, the designer can tune the harvester dimensions and resistive load to obtain the maximum power transfer from the harvester to the conditioning circuit. Different circuits can be also tested to achieve the desired performance requirements.
Figure 3: Cadence Virtuoso circuit schematic and Spectre analysis, with a graph showing power transferred to the bridge load resistor as a function of resistance value
3. Verification and final analysis with FEA
Further detailed modeling can be completed using CoventorWare’s field solvers. For example, the design can be reviewed for high stress areas that may lead to overload failures due to mechanical shock. Gas damping coefficients can also be obtained with CoventorWare and included in the MEMS+ model to more accurately predict Q-factor. When necessary, simulation results from MEMS+ and CoventorWare can be verified against each other to provide increased confidence prior to tape-out. For instance, the closed-loop harmonic response with a linear resistive load can be simulated in both tools and compared.
Figure 4: Simulation in CoventorWare of gas damping forces acting on the vibrating harvester
A Complete Platform
Coventor’s platform for piezoelectric energy harvesting combines MEMS+ and CoventorWare to provide a complete design solution. The CoventorWare platform solves the coupled and multi-domain physics problems not addressed with conventional FEA analysis alone. This hybrid methodology has many advantages. It allows the co-design and co-simulation of an energy harvesting device together with the conditioning circuit. The models are parametric and simulate quickly, making rapid exploration of design and process changes practical. The time-consuming process of creating reduced-order models from FEA data and/or analytical expressions is eliminated, and the resulting models are highly accurate.
In addition, a platform that integrates best-in-class simulators (like Cadence Spectre and/or Matlab/Simulink) with Coventor’s MEMS-specific models provides an optimal combination of accuracy and capacity. Cadence and Mathworks will continue to include speed and capacity improvements in their platform, while CoventorMP continues to provide additional (and improved) higher-order elements in our MEMS+ library. Individual improvements from each of these companies provide MEMS designers with a highly-integrated and productive MEMS design platform.